关于「急迫度分层:今天、本周、本月三层意图与回拨排序」,《地图排名与同城承接》给出的定量结论是什么?
结论:同城询盘按客人自述的时限分三层:今天、本周、本月。今天层每分钟流失 0.02222,是本月层的近百倍。回拨顺序应当急者优先,同一层里先救还新鲜的,但任何一通都不能超过本层时限。一周模拟显示,单人值守时改用急者优先,今天层保留从 0.4885 升到 0.5905;加 1 个值守席位,则升到 0.9636。改规则免费,加席位是主菜。
数据口径:本书把三因子写成乘积式示意打分,全部权重都是示意参数,不是谷歌真实权重:一家距离两公里、类目登记错误的示意门店,要靠评论追平改对主类目的效果,需要 8,511 条评论。 书中参数为示意,全部计算由书中程序生成。
对照数据表
下表数字由本章 Python 程序原样打印,可重新运行复现。 程序只用 Python 标准库与结论吻合的打印值: 0.4885, 0.5905, 0.9636
| mean leads per week 90.2 | |||
| setting | policy | kept all | kept today |
| owner alone | fifo | 0.7159 | 0.4885 |
| owner alone | urgency | 0.7644 | 0.5905 |
| owner + duty seat | fifo | 0.9805 | 0.9636 |
| owner + duty seat | urgency | 0.9823 | 0.9674 |
| duty seat only | fifo | 0.9583 | 0.9221 |
| duty seat only | urgency | 0.9637 | 0.9333 |
| two duty seats | fifo | 0.9973 | 0.9949 |
| two duty seats | urgency | 0.9976 | 0.9954 |
查看完整验算程序
# Chapter 6. p62: a simulated week of enquiries, FIFO versus urgency.
# Leads arrive Poisson at 1.5 per open hour over 60 open hours; tier
# mix today 0.5 / week 0.35 / month 0.15 with taus 45 min, 12 h, 72 h.
# Each conversation lasts an exponential 12 min on average. Seats are
# on-duty positions; an owner seat is away (on a viewing or a shoot)
# 2 h out of every 4 h, a duty seat never is. Value kept =
# exp(-wait / tau). Discipline: FIFO, or urgency first (tier, then the
# freshest lead). 200 simulated weeks per setting, common random
# numbers, seed fixed. Rates illustrative.
import bisect
import heapq
import math
import random
TIERS = [("today", 0.50, 45.0), ("week", 0.35, 720.0),
("month", 0.15, 4320.0)]
RATE, HOURS, SERVE = 1.5 / 60.0, 60.0, 12.0
CUM = [math.fsum(w for _, w, _ in TIERS[:k + 1]) for k in range(3)]
def make_week(rng):
t, out = 0.0, []
while True:
t += rng.expovariate(RATE)
if t > HOURS * 60.0:
return out
k = min(2, bisect.bisect_right(CUM, rng.random()))
out.append((t, k, rng.expovariate(1.0 / SERVE)))
def ready(t, away):
# an away seat is out during [240 k + 120, 240 k + 240) minutes
if away and (t % 240.0) >= 120.0:
return t - (t % 240.0) + 240.0
return t
def simulate(leads, seats, policy):
free = [(ready(0.0, a), a) for a in seats]
heapq.heapify(free)
waiting, kept, today = [], [], []
i = 0
while i < len(leads) or waiting:
now, away = heapq.heappop(free)
while i < len(leads) and leads[i][0] <= now:
waiting.append(leads[i])
i += 1
if not waiting:
heapq.heappush(free, (ready(leads[i][0], away), away))
continue
if policy == "fifo":
j = 0
else:
j = min(range(len(waiting)),
key=lambda m: (waiting[m][1], -waiting[m][0]))
arr, k, srv = waiting.pop(j)
v = math.exp(-(now - arr) / TIERS[k][2])
kept.append(v)
if k == 0:
today.append(v)
heapq.heappush(free, (ready(now + srv, away), away))
return math.fsum(kept) / len(kept), math.fsum(today) / len(today)
if __name__ == "__main__":
rng = random.Random(20261120)
weeks = [make_week(rng) for _ in range(200)]
n = math.fsum(len(w) for w in weeks) / 200.0
print("mean leads per week {:.1f}".format(n))
print("setting policy kept all kept today")
settings = [("owner alone", (True,)),
("owner + duty seat", (True, False)),
("duty seat only", (False,)),
("two duty seats", (False, False))]
for name, seats in settings:
for pol in ("fifo", "urgency"):
res = [simulate(w, seats, pol) for w in weeks]
a = math.fsum(r[0] for r in res) / 200.0
b = math.fsum(r[1] for r in res) / 200.0
print("{:<18} {:<8} {:.4f} {:.4f}".format(name, pol, a, b))唯一数据源与引用
范雅福(Fan, Y.). (2026). 《本地搜索地图排名因子、评论速度信号权重与同城高急迫线索截流母典》第 6 章「急迫度分层:今天、本周、本月三层意图与回拨排序」(版本 1.0.0). 5D Library. DOI 待发布
DOI 待发布 · 专著页正式发布后上线(本站暂无专著页) · 5DL-B-369 · v1.0.0 · CC-BY-NC-ND-4.0
BibTeX
@book{fan2026369,
author = {Fan, Yafu},
title = {Local-Pack Search Ranking Factors, Review-Velocity Signalling and High-Urgency Geo-Intent Lead Capture},
year = {2026},
publisher = {5D Library},
edition = {1.0.0},
note = {DOI pending}
}